ENERGY FORECAST IN A LARGE PV PLANT UNDER PARTIAL SHADING BY KRIGING-BASED APPROACH

In this paper, the use of the kriging estimation method for the study of partial shading in photovoltaic fields is investigated. Different cases of shadowing spatial distribution on the PV field surface, due to different relative positions between the plant and some clouds, are studied. The adopted method gives a spatial interpolation of the shadowing so that its distribution over the entire surface can be obtained on the basis of a limited amount of experimental data. In all cases, the kriging estimates are nearly the same as the observed data, even when very few experimental observations are available. The accuracy of the estimates is also confirmed by Q-1 and double kriging cross-validation schemes. The followed approach allows a cheaper and simpler characterization of the PV plant output power, allowing energy forecasting.